Predicting the spread of marine species introduced by global shipping

Hanno Seebensa,b,c,1, Nicole Schwartza, Peter J. Schuppa, and Bernd Blasiusa

aInstitute for Chemistry and Biology of the Marine Environment, University of Oldenburg, 26111 Oldenburg, Germany; bDepartment of Botany and Biodiversity Research, University of Vienna, 1030 Vienna, Austria; and cSenckenberg Biodiversity and Climate Research Centre, 60325 Frankfurt, Germany

Edited by Alan Hastings, University of California, Davis, CA, and approved March 11, 2016 (received for review December 11, 2015)

The human-mediated translocation of species poses a distinct threat quality data have been made accessible by various online databases, to nature, human health, and economy. Although existing models but testing model predictions with these data has still been lacking. calculate the invasion probability of any species, frameworks for Model frameworks to predict the likelihood of new invasions species-specific forecasts are still missing. Here, we developed a model have already been developed (10, 11, 13). However, these were not approach using global ship movements and environmental conditions able to predict the identity of new invaders, but only the likelihood to simulate the successive global spread of marine alien species that that any new species arrives from a certain source region on Earth. allows predicting the identity of those species likely to arrive next in a Here, we combined such a model, a slightly modified version of the given habitat. In a first step, we simulated the historical stepping- vector-based model of marine invasion adopted from ref. 10, with datasets about the global distribution of marine alien species, which stone spreading dynamics of 40 marine alien species and compared enabled us to predict the identity of the next species to arrive in a predictedandobservedalienspeciesranges.Withanaccuracyof given local habitat. The model is a statistical model that describes 77%, the model correctly predicted the presence/absence of an alien how the probability a given species successfully invades a specific species in an ecoregion. Spreading dynamics followed a common location depends on the shipping traffic and the environmental pattern with an initial invasion of most suitable habitats worldwide differences (temperature and salinity) between locations. and a subsequent spread into neighboring habitats. In a second step, In a first step, to test the accuracy of the model, we used native we used the reported distribution of 97 marine algal species with a ranges of 40 marine species from various taxonomic groups, known invasion history, and six species causing harmful algal blooms, ranging from algae to fish, as initial condition. For each species, we to determine the ecoregions most likely to be invaded next under simulated the global spread outside its native range and compared climate warming. Cluster analysis revealed that species can be the predicted alien range at each simulation time step with the classified according to three characteristic spreading profiles: emerg- observed one. This procedure allows the assessment of the quality ing species, high-risk species, and widespread species. For the North of model predictions, although the degree of expansion distinctly Sea, the model predictions could be confirmed because two of the varied among species, as some species are already widespread, predicted high-risk species have recently invaded the North Sea. This whereas others occupy only a few alien regions either because they study highlights that even simple models considering only shipping just started to spread or there are only a limited number of suitable intensities and habitat matches are able to correctly predict the habitats available. identity of the next invading marine species. In a second step, we used the reported distribution of 97 marine algal species with a known invasion history and six harmful algal alien species | model predictions | species identity | marine ecoregion | species obtained from AlgaeBase (www.algaebase.org)andde- termined the species-specific invasion probabilities for each marine climate change ecoregion not occupied by that species. Algae are particularly well suited for such an analysis because they are easily translocated by the he number of alien species transported by human assistance has exchange of ballast water, and especially invasive seaweeds are of Tincreased rapidly during the last decades with serious conse- global concerns because over 400 introductions have been reported quences for native flora and fauna (1–3). These biological invasions are considered to be one of the major drivers of biodiversity changes Significance (4–6). Once an unwanted alien species has naturalized in the new environment, it is nearly impossible to eradicate the species, and Predicting the arrival of alien species remains a big challenge, thus the mitigation of further introduction is the most efficient way which is assumed to be a consequence of the complexity of the of combating biological invasions (6, 7). However, a targeted mon- invasion process. Here, we demonstrate that spreading of alien itoring and an efficient adaptive management requires knowledge marine species can be predicted by a simple model using only about spreading dynamics of the next potential invaders and thus global shipping intensities, environmental variables, and species about the distribution of species, their invasiveness, and the likeli- occurrence data. We provide species lists of the next potentially hood of new introductions. Although all of these topics have been invading species in a local habitat or species causing harmful algal analyzed on their own, the potential to predict the spreading of alien blooms with their associated probability of invasion. This will species while combining these components remains to be tested. help to improve mitigation strategies to reduce the further in- A large amount of recent introductions can be attributed to the troduction of alien species. Although this study focuses on ma- intensified global trade and transport as many species were acci- rine algae, the model approach can be easily adopted to other dentally or deliberately translocated through the exchange of taxonomic groups and their respective drivers of invasion. commodities or the movements of transportation means (8, 9). The amount of exchanged commodities and the intensity of global Author contributions: H.S., N.S., P.J.S., and B.B. designed research; H.S. and N.S. performed traffic have therefore been foundtobeagoodpredictortomodel research; H.S. and B.B. analyzed data; and H.S., N.S., P.J.S., and B.B. wrote the paper. the global spread of alien species (10–12).Inmostcases,predictions The authors declare no conflict of interest. of alien species introductions are difficult to assess as model results could not be validated (i.e., quantitatively assessed using observed This article is a PNAS Direct Submission. data) thoroughly due to the paucity of high-quality distributional Freely available online through the PNAS open access option. data of alien species. Without any model validation, however, it is See Commentary on page 5470. nearly impossible to assess the quality and the reliability of model 1To whom correspondence should be addressed. Email: [email protected]. predictions, which hampers the application of models for the This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. management of alien species. In recent years, appropriate high- 1073/pnas.1524427113/-/DCSupplemental.

5646–5651 | PNAS | May 17, 2016 | vol. 113 | no. 20 www.pnas.org/cgi/doi/10.1073/pnas.1524427113 Downloaded by guest on September 26, 2021 A invasion probabilities to unoccupied ecoregions. These species 70 SEE COMMENTARY Numbers of observed species Numbers of observed mostly live in temperate to subtropical marine ecoregions in 60 Europe and East Asia (Fig. 1A), corresponding to the known 50 global hot spots of algal diversity (16, 17). The predicted invasion probabilities combined for all of these species reflect the global 40 shipping intensity as marine ecoregions with intense ship traffic 30 such as East Asian Seas or Northern European Seas have a high 20 invasion probability (Fig. 1B). The East Asian Seas and the 10 Northern European Seas are characterized by similar environ- mental conditions and are well connected by ships, which in- creases the likelihood of exchanges between both regions. B >0.20 Indeed, the reported native and alien ranges show that species

Probability of invasion have been mutually exchanged between both regions as many 0.15 alien species in the Northern European Seas originate from East Asia (n = 11 of 19) and vice versa (4 of 16). Our model calcu- 0.10 lations revealed rankings of high-risk algal species for each ecoregion, which is exemplarily shown for the North Sea (Table

0.05 S2). Two of the algal species of the top 10 high-risk species for the North Sea are known from the literature to have naturalized

0.00 in that ecoregion in recent years [Prorocentrum minimum (18) and Polysiphonia harveyi (19)]. These species are not listed for the North Sea in AlgaeBase probably due to delays in data ac- Fig. 1. Observed cumulative distribution (native plus alien) of 97 marine algal species with an invasion history (A) and predicted invasion probabilities for quisition. That is, based on data provided by AlgaeBase, the further spread of these species to formerly unoccupied marine ecoregions. model predicts a high probability that these species will enter the Colors indicate the number of reported algal species in A and the predicted North Sea, which could indeed be confirmed by recent studies. probability of invasion in B, respectively. Predicted climate warming will likely modify the similarity of environmental conditions of marine ecoregions and thus will affect the probabilities of invasion. Using predicted mean sea surface worldwide (14). Furthermore, seaweeds deeply shape marine eco- temperatures during 2040–2060 revealed a latitudinal gradient in systems and they can have strong detrimental ecological and the change of invasion probabilities (Fig. 2): in general, the model economic impacts (14, 15). Using our model, we identified the likely predicts a decrease in invasion probabilities in the tropics, and an hot spots of future invasions among 90 marine ecoregions of the increase in temperate regions due to climate change. Highest in- world and algal species with the highest probability to arrive next. creases in invasion probabilities can be expected for the Northeast Pacific and the Baltic Sea, whereas highest declines arise for Results ecoregions with high shipping intensity in the tropics such as Comparing observed and predicted alien ranges of 40 marine spe- ecoregions with access to the Panama Canal, the Persian Gulf, the cies reveals that the model correctly predicted the presence/absence Strait of Malacca, and South China. of the species in an originally alien ecoregion in 77% of all cases The invasion probabilities distinctly vary among algal species, (median of all 40 spread simulations; Fig. S1). This value is robust to which—in our model—mainly depends on the species-specific interspecific variation (e.g., using only 50% of the species reduced global distribution of algae and their occupancy of ecoregions with the number of correctly predicted ecoregions only to a median of high shipping intensity (thereby enhancing the chance of further 75%; Fig. S2), model parameterization, model structure, and vari- spread). We found that the total invasion probability Ptot(Inv)(i.e., ation of shipping intensity (SI Text and Table S1). Compared with the probability for a species to invade any unoccupied ecoregion observational data, the model overpredicts the number of alien worldwide) is a hump-shaped function of the initial number of oc- species per ecoregion (Fig. S1). Predicted alien species numbers cupied marine ecoregions (Fig. 3A). This can be explained by two were distinctly higher for Southern Africa and East Asia, which are contrasting mechanisms: increasing the number of occupied ecoregions less studied regions, and thus higher alien species numbers can be results (i) in an increase in propagule pressure to unoccupied expected, but also for the US West Coast. ecoregions, thereby increasing the likelihood of further spread, and The simulations of the global spreading dynamics of the 40 (ii) in a reduction of the number of potential new habitats, which marine species were used to analyze common patterns of the global simultaneously decreases the chance of further spread. Combining spread. For each simulated invasion step of each species, we de- both relationships results in the described hump-shaped curve. termined the great circle distance from the centroids of the newly invaded ecoregion to the nearest already occupied ecoregion. This

shows that long-distance jumps are very common among the very Change of invasion probabilities first invasion steps (Fig. S3). For 44% of all species spreads, the 0.05 longest jump to a new ecoregion was found among the first three invasion steps with 10,376 km on average. That is, spreading dy- namics of these species follow a common pattern: first, the suitable 0.00 habitats worldwide were invaded irrespective of the distance from where the species originated, and subsequently the species spread −0.05 to neighboring habitats. This results in a mean geographic distance between observed native and alien ranges of around 10,200 km. Alien species numbers weighted by their number of native and −0.10 alien regions decreased to shorter and longer distances (Fig. S4). A similar although more complex pattern was also found in ref. 10, Fig. 2. Predicted changes in invasion probabilities due to climate warming. which indicates that the models applied in both studies revealed To simulate climate warming, invasion probabilities for 97 algal species with similar results (Materials and Methods). a known invasion history were calculated using water temperatures during

After validation of model predictions, we used the distribution of 2040–2060 for the recipient ecoregions. The resulting invasion probabilities ECOLOGY 97 algal species with a known invasion history obtained from were compared with those calculated using current water temperatures, and AlgaeBase as initial conditions to determine the species-specific the deviations are indicated by colors.

Seebens et al. PNAS | May 17, 2016 | vol. 113 | no. 20 | 5647 Downloaded by guest on September 26, 2021 Consequently, species with narrow or wide distributions have a their native region (i.e., to establish populations viable for years), comparatively low total invasion probability, whereas species with they may produce harmful algal blooms in these regions at least an intermediate distribution have the highest invasion probabilities. during one season of favorable conditions. We therefore selected In addition to the number of occupied ecoregions, the interspecific the distributions of six algal species from AlgaeBase responsible for variation of Ptot(Inv) can be further explained by the maximum either well-known human diseases such as ciguatera or various invasion probability Pmax(Inv) to any ecoregion (Fig. 3B). shellfish poisonings (24) and calculated invasion probabilities for To characterize the spreading potential of each species, we in- each species. Most of these species have an intermediate to wide troduced an invasion threshold « and counted the number of distribution and a comparatively high Ptot(Inv)(Fig.4).Karenia ecoregions with an invasion probability above this threshold. Note brevis, for example, causing neurotoxic shellfish poisoning, has a thatadecreaseof« can be interpreted as a general increase in the Ptot(Inv) of 0.105, which is in the range of the highest invasion probability of invasion due to e.g., elevated propagule pressure, probabilities calculated for species with a known invasion history. interspecific variations in the ability to invade another ecoregion, The reason is that the K. brevis was found in an intermediate or increased environmental match due to environmental changes. number of ecoregions with intense ship traffic (Fig. 4). According Decreasing « (or increasing the invasion probability) leads to a to our classification system for species with known invasion histo- higher number of ecoregions that potentially can be invaded by ries (Fig. 3), Gambierdiscus toxicus would fall in the category of the species (Fig. 3C). The obtained species-specific invasion curves emerging species, whereas Protoperidinium crassipes, Dinophysis saturate when the species occupy all initially unoccupied ecoregions acuminata,andKarenia brevis could be classified as high-risk spe- worldwide. For example, species with a wide initial distribution can cies, and the remaining algae would be widespread species. only invade a few remaining unoccupied ecoregions, resulting in a low number of ecoregions at which the invasion curves saturate. Discussion Cluster analysis reveals three characteristic groups of invasion The mitigation of further introductions of alien species is a chal- curves: the first group, which we call the “emerging species,” has a lenging task mostly due to the complexity of the invasion process narrow initial distribution (red dots in Fig. 3A and red lines in Fig. and the lack of data. Models can help to improve mitigation 3 C and D). A decrease in the invasion threshold « results in a slow strategies by the determination of hot spots of biological invasions, increase of the number of unoccupied ecoregions with an invasion the determination of high-risk pathways, and the identification of threshold above «. That is, the species have a low chance to invade species likely to arrive next. Several studies have tried to predict the another ecoregion. Species of the second group, called the “high- likelihood of new invasions (10, 13, 25–28); however, the common risk species,” have an intermediate initial distribution (green dots paucity of model validations hampers the assessment of the quality in Fig. 3A). Their spreading potential increases comparatively of model predictions and their comparison between different ap- steeply even at high invasion thresholds (green lines in Fig. 3 C proaches. Here, we provide a robust approach for the validation of and D), and thus these species have a high chance to invade other colonization models, simulating the spread of alien species, which ecoregions. The third group of species, called the “widespread may serve as an example for other models and taxonomic groups. species,” consists of species with a wide initial distribution (black We show that our model correctly predicts the invasion status of dots in Fig. 3A). Their invasion curves increase with decreasing species in ecoregions outside their native range in good agreement invasion threshold at a comparatively low rate similar to that of with observations (77% of the presence/absence of an alien species the emerging species (black lines in Fig. 3 C and D), but saturate correctly predicted). This is particularly striking because of the at a low number of unoccupied ecoregions simply due to the low simplicity of the model, considering only habitat matches and availability of ecoregion without the species. maritime traffic intensity and lacking seemingly important predic- Although the known invasion of a species in an ecoregion is the tors like species-specific characteristics, biotic interactions, addi- best predictor for the potential invasiveness of that species for tional environmental variables, or historic shipping data. Indeed, other ecoregions (20, 21), it is interesting to assess the spreading the consideration of additional parameters in the model such as potentials of species, which are not known for their invasiveness further environmental variables did not improve the model fit but may affect other species and potentially even ecosystems. Some (Table S1). The quality of the fit ranges between those found in algae produce toxic substances causing fish or shellfish poisoning, other modeling studies [e.g., R2 = 0.64 for the global spread of which can also be harmful for humans (22). Transportation of toxic terrestrialvascularplants(29)oran87–94% accuracy for the in- algal cells or cysts in ballast water of ships is a likely explanation for troduction of fishes into the Great Lakes (27)], but other models harmful algal blooms in previously unaffected regions (22, 23). were either more complex or restricted to much smaller geographic Although these species may not be known to naturalize outside areas or a single taxonomic group. We are not aware of any study,

A 0.10 C 80 emerging species Fig. 3. Spreading potential of marine alien algae. emerging species high risk species widespread species 0.08 high risk species 60 (A) The total invasion probabilities Ptot(Inv) of algal  widespread species species follow a hump-shaped curve of the number

Inv 0.06  40 of initially occupied marine ecoregions, with species tot

P 0.04 of intermediate global distribution having the highest P (Inv). The colors denote the category of spreading

Number of newly 20 tot 0.02 occupied ecoregions potentials classified by cluster analysis (C and D). 0.00 0 (B) The same data as in A, but the colors indicate the

maximum invasion probability Pmax(Inv) from the B D occupied range of the species to any unoccupied 0.10 P maxInv 80 4 ecoregion. (C) Invasion curves for each species esti- 2 mated as the number of ecoregions outside the range, 0.08 0.02 60 Ecoregions  0 3 2 1 which the species initially occupied, above an invasion 10 10 10 « Inv 0.06 Invasion threshold  threshold . Cluster analysis of interspecific variations  40 tot identified three groups of species with different in-

P 0.04 0.01 vasion curves (red, emerging species; green, high-risk

Number of newly 20 0.02

occupied ecoregions species; and black, widespread species). (D)Thesame as C, but showing the mean invasion curves for each 0.00 0.00 0 0 10203040506070 109 107 105 103 101 cluster. The Inset indicates a zoom to high invasion Number of occupied ecoregions Invasion threshold  threshold values.

5648 | www.pnas.org/cgi/doi/10.1073/pnas.1524427113 Seebens et al. Downloaded by guest on September 26, 2021 SEE COMMENTARY

Fig. 4. Invasion probabilities for a selection of six algal species causing fish and shellfish poisonings potentially harmful to humans. The selected algae are known to produce toxic compounds causing the respective poisoning of fish or shellfish. Black areas denote the current distribution of species, whereas colors indicate the invasion probability to an un- occupied ecoregion. The total invasion probabilities

Ptot(Inv) expressed as the integration of all single invasion probabilities are given in the subheading of each panel.

simulating the global spread of marine species from various taxo- (10), and an elevation of water temperatures in the Northeast nomic groups simultaneously as done here, which renders a direct Pacific will increase the environmental match of both regions, comparison difficult. Probably the best indication of the model’s resulting in increased invasion probabilities. Indeed, a recent quality is the fact that two of the algal species predicted to be finding of the Asian macroalga Sargassum horneri at the West among the top 10 high-risk species for the North Sea are known to Coast of Mexico and the United States is likely is a consequence have naturalized in that ecoregion in recent years. That is, in two of this increased environmental match (30), which further sup- cases our species-specific predictions could be confirmed by recent ports our model predictions. samplings, which are not yet included in AlgaeBase. This also Our model predicts the total invasion probability of a species to shows that it is possible to explain the spread of alien species to a large degree by simple vector-based invasion models. The sensi- be a hump-shaped function of the initially occupied number of tivity analysis (SI Text) highlights that our model predictions are ecoregions (Fig. 3 A and B), the reason being that species with an robust to the choice of species for model validation, and to the intermediate global distribution pose a large propagule pressure to variation in model structure, parameterization, and input datasets. a comparatively large number of unoccupied ecoregions. This For the 97 marine algal species considered in this study, the pattern corresponds to the relationship of the colonization rate on highest probability of invasion arises in Asian and European the number of occupied patches of classical metacommunities ecoregions (Fig. 1B), mostly along the major shipping routes. The models (31). In these models, the colonization rate of new patches Asian Seas were also identified as invasion hot spots in a previous has to be zero if either none or all patches are occupied by a analysis of marine invasion, thereby ignoring observed species species and positive in between, which indicates the close re- distributions (10). In contrast to our study, the invasion proba- lationship between both types of models. bilities of Northern European Seas were very low in ref. 10, which The reliable prediction of invasion dynamics remains one of the was explained by the, on average, low environmental similarity to biggest challenges in invasion ecology. We here demonstrate an most other ecoregions worldwide. The reason for the elevated approach for a robust validation of global invasion models, which invasion probabilities found in this study is that many algal species allows the assessment of the quality of model predictions. This predicted to be at high risk in Northern European Seas are native to East Asian waters, which provide similar environmental con- study highlights that the combination of invasion models with ditions to Northern European Seas. As both regions are highly observational data can essentially improve the predictions of in- interconnected due to intense ship traffic, the probability to invade vasion probabilities. The model applied here can be easily adopted from one region to the other is also high, resulting in the described to simulate the spread of other taxonomic groups (see ref. 29 for mutual exchange of species between these regions. This shows an example of terrestrial vascular plants). Many online databases that, although the overall patterns of both studies were similar, emerged in recent years. Notable examples are the Delivering observational data are important to further improve the predic- Alien Invasive Species Inventories for Europe (DAISIE) (www. tions of invasion hot spots. europe-aliens.org), the Invasive Species Compendium by the Our model predicts that climate warming will lead to reduced Centre for Agriculture and Bioscience International (CABI), or invasion probabilities in the tropics and elevated ones in temperate the upcoming Global Register of Invasive Species (GRIS) by the regions, particularly in North America (Fig. 2). The reason is that Invasive Species Specialists Group (www.issg.org), all of which many species used in this study live in temperate to subtropical provide native and alien ranges of numerous species. It is there- ecoregions and were assumed to be adapted to water temperatures of those regions. If water temperatures increase, these species will fore the logical consequence to combine invasion models with find appropriate environmental conditions at higher latitudes than available distributional data to improve the predictability of in- inhabited now, whereas environmental conditions will get less vasion dynamics as shown here. The prediction of future invasions

suitable in tropical regions. The highest increase in invasion is a prerequisite of efficient mitigation strategies, and the de- ECOLOGY probability arises for the Northeast Pacific. For this ecoregion, the termination of the species identity enables a targeted monitoring by-far most important donor area constitutes the Northwest Pacific of potential high-risk species.

Seebens et al. PNAS | May 17, 2016 | vol. 113 | no. 20 | 5649 Downloaded by guest on September 26, 2021 Materials and Methods parameterization, the input variables, and tests of the robustness of this Data. As input variables, the model requires data about the arrival and de- approach are provided in ref. 10. parture dates of single ships moving between ports, ship type, ship size, ballast tank volumes, and environmental data of ports. The ship and route-specific data Model Validation. For model validation (i.e., the quantitative assessment of the (i.e., ship size, ship type, and arrival and departure dates) were obtained from a quality of model predictions using observed data), we compiled data of native large dataset of ship movements and ship characteristics provided by Lloyd’s and alien distributions of marine alien species from the CABI Invasive Species Register Fairplay (www.ihs.com), consisting of nearly 3 million ship voyages of Compendium (www.cabi.org/isc/). In a first step, we used the implemented “ ” 32,511 ships between 1,469 ports during 2007–2008 (10), and ballast tank search engine of CABI with the search term (HAB marine) AND ballast water volume data taken from the American Bureau of Shipping (32). The shipping to filter for species in marine habitats likely to be transported by ballast water. intensity was assumed to be constant in time (but see the discussion about the We manually removed species not fitting to this category as, e.g., freshwater reliability of using recent ship movement data to predict historic spreading species with a high salinity tolerance were included as well, with the exception dynamics in SI Text). Sea surface temperatures and nutrient concentrations of Dreissena polymorpha as this species is known to be transported by open-sea (nitrate, phosphate, and silicate) were obtained from the World Ocean Atlas vessels (34). We also removed regions with uncertain invasion status of a spe- (WOA) (www.nodc.noaa.gov), providing 50-y averages at 1° spatial resolution. cies, which resulted in a total of 40 species with known native and alien dis- Surface salinity data of ports were calculated from port-specific data of water tributions from various taxonomic groups ranging from algae to fish. The densities provided by Lloyd’s Register Fairplay (www.sea-web.com/portguide. following species were used for model validation: Acanthogobius flavimanus, html) for most of the ports (69%). Water densities were recalculated to salinities Acentrogobius pflaumii, Ascidiella aspersa, Asterias amurensis, Austrominius using temperatures taken from the WOA. For the remaining ports, salinities modestus, Caprella mutica, Carcinus maenas, Caulerpa racemosa var. cylin- were taken from the WOA. Future sea surface temperatures were obtained from dracea, Charybdis hellerii, Ciona intestinalis, Ciona savignyi, Crassostrea virgin- the Coupled Model Intercomparison Project (CMIP5) (cmip-pcmdi.llnl.gov/cmip5/) ica, Crepidula fornicata, Diplosoma listerianum, Dreissena polymorpha, Ensis providing predictions of sea surface temperatures using an Intergov- directus, Gracilaria salicornia, Grateloupia turuturu, Hemigrapsus sanguineus, ernmental Panel on Climate Change scenario of intermediate greenhouse Hemigrapsus takanoi, Hemimysis anomala, Littorina littorea, Marenzelleria gas concentration (RCP6) (33). The average sea surface temperatures were neglecta, Microcosmus squamiger, Mnemiopsis leidyi, Musculista senhousia, calculated for each marine ecoregion for 2040–2060. Mytilus galloprovincialis, Palaemon elegans, Palaemon macrodactylus, Phyllo- rhiza punctata, Polyandrocarpa zorritensis, Pseudochattonella verruculosa, Model. We applied a vector-based model of marine invasion to calculate the Pterois volitans, Rapana venosa, Rhithropanopeus harrisii, Schizoporella errata, probability of invasion by global shipping between 1,469 major ports worldwide Spartina alterniflora, Styela plicata, Ulva pertusa,andUlva reticulate. CABI (10). In the original form, the model consisted of three independent probabili- provides distributions usually on a country basis, but in some cases also sub- ties: the probability to be alien, P(Alien); the probability of introduction, P(Intro); national units such as states, provinces, or islands are provided. We translated and the probability of establishment, P(Estab). Here, however, when we combine the distributional data provided by CABI into marine ecoregions, thereby as- the model with observational data of species distributions, the term P(Alien), suming that if a species was listed for a geographic unit bordering or located in constituting a theoretical estimator for biogeographical dissimilarity is not re- a certain ecoregion, the species can be found in the whole ecoregion including quired anymore. We therefore removed P(Alien) from the model. the ports. This may result in an overprediction of the actual distribution of ’ According to ref. 10, the probability of introduction, species, but given the highly patchy sampling coverage of the world s coast- lines, simulations can only be done on a coarse geographic resolution, which −λ −Δ Br tr necessitates such simplifications. A marine ecoregion represents an area of Pr ðIntroÞ = 1 − e e , [1] similar environmental conditions and species assemblages, and thus it seems

is a function of exchanged ballast water volume Br on route r between donor likely that a species can also be found in other parts of that ecoregion. We port i and recipient port j and travel time Δt. Pr(Intro) is obtained for each adopted the classification of “marine ecoregions of the world” by Spalding single ship on a certain route r taken from the ship movement dataset. Br is et al. (35) but considered only those ecoregions that have at least one port δr calculated as Br = zWr ð1 − zWr =Vr Þ , with Wr being the ship type- and ship listed in our shipping database, resulting in a total of 90 marine ecoregions. To size-specific amount of released ballast water in cubic meters obtained from assess the reliability of using recent ship movement data for the prediction of

regression fits shown in figure S1 in ref. 10, Vr denoting the ship size-specific historic spreading dynamics of alien species, we compiled a list of the years of mean volume of ballast tanks of a ship, δr representing the number of stop- first record of the alien species in a region (Table S3). Inevitably, the year of first over ports on route r,andz being the fraction of nonzero releases record is not provided for all species and all regions due to the lack of data. depending on the type and the size of a ship. Note that in ref. 10, z was For each species, we calculated the mean temperature and salinity require- falsely described as the fraction of zero releases, although it has to be the ments from their known native distribution by taking the average of all native fraction of nonzero releases as stated here. ecoregion means of temperature and salinity. The obtained species-specific

The probability of establishment, ecological niche was used as input variables Ti and Si in Eq. 2. Note that i now represents the geographic area used to calculate the ecological niche of a h i

Δ Δ 1 = Tij Sij species. As initial condition, we set all ports where the species is considered to − 2 σ + σ T S Pij ðEstabÞ = αe , [2] be native as “occupied” and all other ports to “unoccupied.” We then applied the model to calculate the invasion probability to all unoccupied ports. The port Δ was modeled as a Gaussian function of differences of water temperature Tij with the highest invasion probability was identified and set to occupied. This Δ and salinity Sij of donor port i and recipient port j standardized by the was repeated until a threshold of a very low invasion probability was reached, σ σ α width of the respective ecological niche T and S. represents the basic which corresponds to a hypothetical nearly worldwide coverage of the species. probability of establishment. To analyze the influence of climate warming The sequence of invaded ports was transformed to the sequence of invaded on model predictions, Pij(Estab) was calculated using future sea surface ecoregions, thereby removing duplicated ecoregions. For each time step of the – temperature (2040 2060; Data) for the recipient regions. modeled spread, we compared the predicted alien distribution of the species The probability of invasion from port i to port j is given by the comple- with the observed alien distribution obtained from CABI and calculated the ment of species failing to invade on all ship routes rij connecting both ports, goodness-of-fit of predicted and observed alien distributions. The goodness-of- fit was estimated by the number of correctly predicted invaded ecoregions (true ð Þ = −∏ − ð Þ ð Þ Pij Inv 1 1 Pr Intro Pij Estab . [3] positives) divided by the sum of false predictions (false positives plus false rij negatives). At the time step of the largest goodness-of-fit value, the deviation The invasion probabilities between ports were aggregated accordingly to between predicted and observed distribution was lowest, and the respective

obtain the invasion probabilities Pj(Inv) from the ecoregions occupied by the distribution was selected as the best fit. This procedure was repeated for each respective species to each unoccupied ecoregion j. The total invasion prob- of the 40 species and the median of the resulting 40 goodness-of-fit values was

ability was calculated for each species as Ptot(Inv) = 1 −∏j[1 − Pj(Inv)] and the taken as the overall goodness-of-fit of the model. maximum probability of invasion as Pmax(Inv) = max(Pj(Inv)). To test the robustness of our model predictions, we performed an ex- Although some modifications of the model structure and parameteriza- tensive sensitivity analysis, thereby varying the simulation process, the model tions slightly improved the accuracy of model predictions (SI Text and Table parameterization, and the distributional data used for model validation, and S1), we adopted the basic model version and the parameterization provided discuss the reliability of the ship movement data for our modeling purpose (SI by ref. 10 for consistency. A more detailed description of the model, its Text).

5650 | www.pnas.org/cgi/doi/10.1073/pnas.1524427113 Seebens et al. Downloaded by guest on September 26, 2021 Model Application. After model validation, we calculated invasion probabilities in the North Sea, we set all North Sea ports to occupied, whereas if the species

using the global distribution of marine algae taken from AlgaeBase (www. was listed for Germany, we selected the North Sea ports and Baltic Sea ports SEE COMMENTARY algaebase.org) to identify future invasion hot spots and high-risk species of located in Germany as occupied ports. For each species, we then calculated the marine algae. We used the species names listed as “accepted names” in the invasion probabilities to any unoccupied port. database. AlgaeBase provides detailed information for numerous algal species, including their current global distribution, but AlgaeBase does not provide the Spreading Potentials of Species. The invasion probabilities obtained from the information whether a species is native or alien. We therefore adopted the model applied to the AlgaeBase data were used to characterize the spreading finding of previous studies that the best forecasting tool to predict the invasion potential of each species. This was done by introducing a threshold e of invasion − − risk is whether the species has naturalized elsewhere (20), which is sometimes probability ranging from 10 9 to 10 1. For each species, the ecoregions outside even taken as the only permanent predictor of invasion success (21). We the species’ native range with an invasion probability above « were counted. screened the literature and other databases to compile a list of algal species This can be interpreted as those ecoregions that can be invaded by the species with a known invasion history. We started with a list of algal species provided given a certain threshold value. While successively reducing «, the number of by Molnar et al. (36) and added other species from the literature and online these ecoregions with an invasion probability above « increased, but at dif- databases and finally ended up with 97 algal species with a known invasion ferent rates depending on the species-specific spreading potentials resulting in history. For these species, we selected their current distribution from AlgaeBase. species-specific invasion curves. To characterize the interspecific differences, we For each species, the ecological niche represented by Ti and Si was calculated as calculated the area bounded between all pairs of invasion curves and applied a the mean temperatures and salinities of all occupied ecoregions. We stan- hierarchical clustering algorithm using the open-source software package R dardized the geographic units to the level of a country, island, or marine (37). The distances between clusters were calculated with “Ward’sminimum ecoregion depending on the level of detail provided by AlgaeBase. For large variance method” (38), which aims at finding compact spherical clusters. We countries (United States, Canada, Russia, and Australia), we distinguish sub- tested other clustering algorithms as well, which, however, did not change the national units such as provinces or states if this information was provided. For results significantly. example, if a species was listed for Australia, we assigned all ecoregions sur- rounding Australia to this species, whereas if the species was found only at the ACKNOWLEDGMENTS. H.S. and B.B. acknowledge financial support from the island of Tasmania or at the coast of Queensland, we only selected the re- Volkswagen Foundation. H.S. was further supported by the Austrian Research spective ecoregions. Assuming that the species is homogeneously distributed in Foundation (FWF) Grant I2096-B16 and by the German Research Foundation the respective geographic unit (along a country coast or in a marine ecoregion), (DFG) Grant SE 1891/2-1, and N.S. by the Research-Oriented Teaching Project all ports in that area were set to occupied. For instance, if the species was found (Forschungsorientierte Lehre) of the University of Oldenburg.

1. Simberloff D, et al. (2013) Impacts of biological invasions: What’s what and the way 26. Paini DR, Yemshanov D (2012) Modelling the arrival of invasive organisms via the forward. Trends Ecol Evol 28(1):58–66. international marine shipping network: A Khapra beetle study. PLoS One 7(9):e44589. 2. Pyšek P, et al. (2012) A global assessment of invasive plant impacts on resident species, 27. Kolar CS, Lodge DM (2002) Ecological predictions and risk assessment for alien fishes communities and ecosystems: The interaction of impact measures, invading species’ in North America. Science 298(5596):1233–1236. traits and environment. Glob Change Biol 18(5):1725–1737. 28. Gallien L, Munkemuller T, Albert CH, Boulangeat I, Thuiller W (2010) Predicting potential 3. McGeoch MA, et al. (2010) Global indicators of biological invasion: Species numbers, distributions of invasive species: Where to go from here? Divers Distrib 16(3):331–342. biodiversity impact and policy responses. Divers Distrib 16(1):95–108. 29. Seebens H, et al. (2015) Global trade will accelerate plant invasions in emerging 4. Butchart SHM, et al. (2010) Global biodiversity: Indicators of recent declines. Science economies under climate change. Glob Change Biol 21(11):4128–4140. 328(5982):1164–1168. 30. Marks LM, et al. (2015) Range expansion of a non-native, invasive macroalga Sargassum 5. Millennium Ecosystem Assessment (2005) Ecosystems and Human Well-Being: horneri (Turner) C. Agardh, 1820 in the eastern Pacific. BioInvasions Rec 4(4):243–248. Biodiversity Synthesis (Island Press, Washington, DC). 31. Levins R (1969) Some demographic and genetic consequences of environmental 6. Pyšek P, Richardson DM (2010) Invasive species, environmental change and man- heterogeneity for biological control. Bull Entomol Soc Am 15:237–240. agement, and health. Annu Rev Environ Resour 35(1):25–55. 32. American Bureau of Shipping (2011) Ballast Water Treatment Advisory (American 7. Puth LM, Post DM (2005) Studying invasion: Have we missed the boat? Ecol Lett 8(7): Bureau of Shipping, Houston). Available at ww2.eagle.org/content/eagle/en.html. 715–721. Accessed October 23, 2012. 8. Levine JM, D’Antonio CM (2003) Forecasting biological invasions with increasing in- 33. Moss R, et al. (2008) Towards New Scenarios for Analysis of Emissions, Climate Change, ternational trade. Conserv Biol 17(1):322–326. Impacts, and Response Strategies (Intergovernmental Panel on Climate Change, Geneva). 9. Hulme PE, et al. (2008) Grasping at the routes of biological invasions: A framework for 34. Ricciardi I, MacIsaac HJ (2000) Recent mass invasion of the North American Great integrating pathways into policy. J Appl Ecol 45(2):403–414. Lakes by Ponto-Caspian species. Trends Ecol Evol 15(2):62–65. 10. Seebens H, Gastner MT, Blasius B (2013) The risk of marine bioinvasion caused by 35. Spalding MD, et al. (2007) Marine ecoregions of the world: A bioregionalization of global shipping. Ecol Lett 16(6):782–790. coastal and shelf areas. Bioscience 57(7):573–583. 11. Keller RP, Drake JM, Drew MB, Lodge DM (2010) Linking environmental conditions 36. Molnar JL, Gamboa RL, Revenga C, Spalding MD (2008) Assessing the global threat of and ship movements to estimate invasive species transport across the global shipping invasive species to marine biodiversity. Front Ecol Environ 6(9):485–492. network. Divers Distrib 17(1):93–102. 37. R Core Team (2014) R: A Language and Environment for Statistical Computing (R 12. Drake JM, Lodge DM (2004) Global hot spots of biological invasions: Evaluating op- Foundation for Statistical Computing, Vienna), Version 3.2.3. Available at www. tions for ballast-water management. Proc Biol Sci 271(1539):575–580. r-project.org. 13. Jerde CL, Lewis MA (2007) Waiting for invasions: A framework for the arrival of 38. Ward JH (1963) Hierarchical grouping to optimize an objective function. J Am Stat nonindigenous species. Am Nat 170(1):1–9. Assoc 58(301):236–244. 14. Williams SL, Smith JE (2007) A global review of the distribution, taxonomy, and im- 39. Hewitt CL, et al. (2004) Introduced and cryptogenic species in Port Phillip Bay, Victoria, pacts of introduced seaweeds. Annu Rev Ecol Evol Syst 38:327–359. Australia. Mar Biol 144(1):183–202. 15. Wallentinus I, Nyberg CD (2007) Introduced marine organisms as habitat modifiers. 40. DAISIE (2009) Handbook of Alien Species in Europe (Springer, Dordrecht, The Neth- Mar Pollut Bull 55(7-9):323–332. erlands). 16. Bolton JJ (1994) Global seaweed diversity: Patterns and anomalies. Bot Mar 37(3): 41. O’Flynn C, Kelly J, Lysaght L (2014) Ireland’s invasive and non-native species—trends 241–246. in introductions. Natl Biodivers Data Cent Ser 2:1–50. 17. Kerswell AP (2006) Global biodiversity patterns of benthic marine algae. Ecology 42. Bungartz F, et al., eds (2009) Charles Darwin Foundation Galapagos Species Checklist— 87(10):2479–2488. Lista de Especies de Galápagos de la Fundación Charles Darwin. Available at www. 18. Hoppenrath M (2004) A revised checklist of planktonic diatoms and dinoflagellates darwinfoundation.org/datazone/checklists/. Accessed November 9, 2015. from Helgoland (North Sea, German Bight). Helgol Mar Res 58(4):243–251. 43. Williams SL (2007) Introduced species in seagrass ecosystems: Status and concerns. 19. Maggs CA, Stegenga H (1998) Red algal exotics on North Sea coasts. Helgol J Exp Mar Biol Ecol 350(1-2):89–110. Meersunters 52(3-4):243–258. 44. Çinar ME, Bilecenoglu˘ M, Öztürk B, Katagan T, Aysel V (2005) Alien species on the 20. Kulhanek SA, Ricciardi A, Leung B (2011) Is invasion history a useful tool for predicting coasts of Turkey. Mediterr Mar Sci 6(2):119–146. the impacts of the world’s worst aquatic invasive species? Ecol Appl 21(1):189–202. 45. Carlton JT, Eldredge LG (2009) Marine bioinvasions of Hawai’i: The introduced and 21. Williamson M (1999) Invasions. Ecography (Cop) 22(1):5–12. cryptogenic marine and estuarine animals and plants of the Hawaiian Archipelago. 22. Hallegraeff GM (1993) A review of harmful algal blooms and their apparent global Bish Museum Bull Cult Environ Stud 4:1–203. increase. Phycologia 32(2):79–99. 46. Coles SL, DeFelice RC, Eldredge LG, Carlton JT (1999) Historical and recent introduc- 23. Van Dolah FM (2000) Marine algal toxins: Origins, health effects, and their increased tions of non-indigenous marine species into Pearl Harbor, Oahu, Hawaiian Islands. occurrence. Environ Health Perspect 108(Suppl 1):133–141. Mar Biol 135(1):147–158. 24. Wang D-Z (2008) Neurotoxins from marine dinoflagellates: A brief review. Mar Drugs 47. Roy HE, et al. (2012) Non-Native Species in Great Britain: Establishment, Detection and 6(2):349–371. Reporting to Inform Effective Decision Making (Non-Native Species Secretariat, York, UK). 25. Herborg LM, Jerde CL, Lodge DM, Ruiz GM, MacIsaac HJ (2007) Predicting invasion 48. Wu S, et al. (2010) Insights of the latest naturalized flora of Taiwan: Change in the ECOLOGY risk using measures of introduction effort and environmental niche models. Ecol Appl past eight years. Taiwania 55(2):139–159. 17(3):663–674. 49. Xu H, et al. (2012) An inventory of invasive alien species in China. NeoBiota 15:1–26.

Seebens et al. PNAS | May 17, 2016 | vol. 113 | no. 20 | 5651 Downloaded by guest on September 26, 2021